Hi,

Thank you for all the interesting links, most of all to the SqueakShell! It is really cool, but however, it actually does not solve my corner.

@Fabio:

...

As far as I found out, Squeak only has a CommandLineToolSet. Am I missing a class? Do you think this would be a useful extension?

Could you elaborate on your use case? Is this simply to be able to write interactive cmd scripts in Squeak?

Exactly. If you run a simple script such as [HelpIcons squeakIcon writeOnFileNamed: 'testicon.png'] in headless mode and testicon.png already exists, the execution will be halted and the VM is waiting for a user interaction that will never happen. It would be much nicer to get a prompt via stdout + stdin to answer.

And even if the VM is run in a read-only shell, you would at least know why it does not continue to execute. For example, this would be helpful on smalltalkCI - if you had not caught all ProvideAnswerNotifications in your tests, you would basically have no hint why you get a timeout.

I know Squeak is primarily made for use in a VM window, but nonetheless, I would love to have an implementation similar to this one for Pharohttps://github.com/StephanEggermont/pharo/blob/development/src/UIManager/CommandLineUIManager.class.st#L104 in Squeak.

Best,

Christoph

________________________________ Von: Squeak-dev squeak-dev-bounces@lists.squeakfoundation.org im Auftrag von tim Rowledge tim@rowledge.org Gesendet: Freitag, 1. November 2019 20:01:50 An: The general-purpose Squeak developers list Betreff: Re: [squeak-dev] CommandLineUIManager

On 2019-11-01, at 11:29 AM, Thiede, Christoph Christoph.Thiede@student.hpi.uni-potsdam.de wrote:

Hi all,

does Squeak already support something like a CommandLineUIManager?

We discussed some stuff around this topic back in May '19 relating to "Running Squeak fro ma unix shell script file with #! squeak..." (including the spilling mistook).

And don't forget CommandShell (http://map.squeak.org/package/2c3b916b-75e2-455b-b25d-eba1bbc94b84), OSProcess (http://map.squeak.org/package/812c9d14-5236-4cad-82ea-cc3e3837e30d), and Eliot's simple exemplar REPL thing (https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/image/LoadReader.... & https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/image/StartReader...)

I also see that http://www.squeaksource.com/@azqxWT7-o5bjyYS2/vd0deS9q claims to include a REPL package (never tried it, see also http://gulik.pbworks.com/w/page/7760030/REPLServer)

And I'm sure Craig has done something related but I simply cannot get google to find it right now

tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Negligent (adj.), describes a condition in which you absentmindedly answer the door in your nightgown.

Hi Christoph,

I don't know if you're actually looking for an interactive REPL, but if all you need is to run a script...

Exactly. If you run a simple script such as [HelpIcons squeakIcon writeOnFileNamed: 'testicon.png'] in headless mode and testicon.png already exists, the execution will be halted and the VM is waiting for a user interaction that will never happen. It would be much nicer to get a prompt via stdout + stdin to answer.

... then you only need to wrap your expression in a Smalltalk run: [ ... ] block. For example, if you wrap your expression and put it into some file, say, "script.st",

"contents of script.st" Smalltalk run: [HelpIcons squeakIcon writeOnFileNamed: 'testicon.png']

then, it'll be ready for headless, you can:

squeak -vm display=none my.image script.st > my.out 2>&1

and it'll redirect all stdout and stderr to file "my.out". All uncaught exceptions (except Halt, I think) are handled and printed.

For servers, one thing I also put at the beginning my run block is:

Smalltalk mitigateIfHeadless

which starts the RFB server, only if it's in headless mode, so I can get in there if I need to.

HTH, Chris

And even if the VM is run in a read-only shell, you would at least know *why* it does not continue to execute. For example, this would be helpful on smalltalkCI - if you had not caught all ProvideAnswerNotifications in your tests, you would basically have no hint why you get a timeout.

I know Squeak is primarily made for use in a VM window, but nonetheless, I would love to have an implementation similar to this one for Pharo https://github.com/StephanEggermont/pharo/blob/development/src/UIManager/CommandLineUIManager.class.st#L104 in Squeak.

Best,

Christoph

*Von:* Squeak-dev squeak-dev-bounces@lists.squeakfoundation.org im Auftrag von tim Rowledge tim@rowledge.org *Gesendet:* Freitag, 1. November 2019 20:01:50 *An:* The general-purpose Squeak developers list *Betreff:* Re: [squeak-dev] CommandLineUIManager

...

On 2019-11-01, at 11:29 AM, Thiede, Christoph <

Christoph.Thiede@student.hpi.uni-potsdam.de> wrote:

...

Hi all,

does Squeak already support something like a CommandLineUIManager?

We discussed some stuff around this topic back in May '19 relating to "Running Squeak fro ma unix shell script file with #! squeak..." (including the spilling mistook).

And don't forget CommandShell ( http://map.squeak.org/package/2c3b916b-75e2-455b-b25d-eba1bbc94b84), OSProcess ( http://map.squeak.org/package/812c9d14-5236-4cad-82ea-cc3e3837e30d), and Eliot's simple exemplar REPL thing ( https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/image/LoadReader.... & https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/image/StartReader... )

I also see that http://www.squeaksource.com/@azqxWT7-o5bjyYS2/vd0deS9q claims to include a REPL package (never tried it, see also http://gulik.pbworks.com/w/page/7760030/REPLServer)

And I'm sure Craig has done something related but I simply cannot get google to find it right now

tim

tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Negligent (adj.), describes a condition in which you absentmindedly answer the door in your nightgown.

Hi Christoph,

unfortunately, I cannot reproduce your command - in my shell, the VM only outputs that "pthread_setschedparam failed: Operation not permitted. This VM uses a separate heartbeat thread ..." warning and then hangs on, even if my script ends by calling Smalltalk quitPrimitive.

I believe it has nothing to do with my command. Even if you try to just run

squeak -version

in that environment, I think you'll get the same error. There are two flavors of the Linux VM, you're using the version of the VM that's a little better, but requires you to install a file into your security limits that allows it to work.

Create a text file called "squeak.conf" with these contents (between the horizontal lines):

* hard rtprio 2 * soft rtprio 2

Copy that file to

/etc/security/limits.d/squeak.conf

and you should now be able to launch the Squeak VM, and my command will work, even with FileExistsException.

---

However, I made the example with #writePNGfileNamed: because if the file exists, it does not raise an Error or some other unhandled exception, but a FileExistsException which, by default, opens a UIManager.

This happens just so often in Squeak that I would like to implement a

CommandLIneUIManager for this case.

You could consider refactoring writePNGfileNamed: and other places to dispatch to an interim-level method that takes the actual Stream you want to write to, that way, instead of having to worry about catching exceptions, you can simply create your own FileStream (with UI's, or not) and use the lower-level method.

Even though I think it could be improved, asking for a file via UI is a real use-case for Squeak, and supporting that use-case in a way that allows overriding via exception handling is a good idea.

Best, Chris

And of course, you might even combine this with the SqueakShell to avoid this inconsistent situation:

If I find some time, I will be happy to upload some proposal into the inbox :)

@Fabio: Yes, separately handling a ProvideAnswerNotification in the SCITestRunner sounds good. However, I would first focus on solving that problem for all CLI applications, but I will keep it in mind :)

Best,

Christoph

*Von:* Squeak-dev squeak-dev-bounces@lists.squeakfoundation.org im Auftrag von Chris Muller asqueaker@gmail.com *Gesendet:* Mittwoch, 6. November 2019 00:27:27 *An:* The general-purpose Squeak developers list *Betreff:* Re: [squeak-dev] CommandLineUIManager

Hi Christoph,

I don't know if you're actually looking for an interactive REPL, but if all you need is to run a script...

...

Exactly. If you run a simple script such as [HelpIcons squeakIcon writeOnFileNamed: 'testicon.png'] in headless mode and testicon.png already exists, the execution will be halted and the VM is waiting for a user interaction that will never happen. It would be much nicer to get a prompt via stdout + stdin to answer.

... then you only need to wrap your expression in a Smalltalk run: [ ... ] block. For example, if you wrap your expression and put it into some file, say, "script.st",

          "contents of script.st"
          Smalltalk run: [HelpIcons squeakIcon writeOnFileNamed:

'testicon.png']

then, it'll be ready for headless, you can:

      squeak -vm display=none my.image script.st > my.out 2>&1

and it'll redirect all stdout and stderr to file "my.out". All uncaught exceptions (except Halt, I think) are handled and printed.

For servers, one thing I also put at the beginning my run block is:

   Smalltalk mitigateIfHeadless

which starts the RFB server, only if it's in headless mode, so I can get in there if I need to.

HTH, Chris

...

And even if the VM is run in a read-only shell, you would at least know *why* it does not continue to execute. For example, this would be helpful on smalltalkCI - if you had not caught all ProvideAnswerNotifications in your tests, you would basically have no hint why you get a timeout.

I know Squeak is primarily made for use in a VM window, but nonetheless, I would love to have an implementation similar to this one for Pharo https://github.com/StephanEggermont/pharo/blob/development/src/UIManager/CommandLineUIManager.class.st#L104 in Squeak.

Best,

Christoph

*Von:* Squeak-dev squeak-dev-bounces@lists.squeakfoundation.org im Auftrag von tim Rowledge tim@rowledge.org *Gesendet:* Freitag, 1. November 2019 20:01:50 *An:* The general-purpose Squeak developers list *Betreff:* Re: [squeak-dev] CommandLineUIManager

...

On 2019-11-01, at 11:29 AM, Thiede, Christoph <

Christoph.Thiede@student.hpi.uni-potsdam.de> wrote:

...

Hi all,

does Squeak already support something like a CommandLineUIManager?

We discussed some stuff around this topic back in May '19 relating to "Running Squeak fro ma unix shell script file with #! squeak..." (including the spilling mistook).

And don't forget CommandShell ( http://map.squeak.org/package/2c3b916b-75e2-455b-b25d-eba1bbc94b84), OSProcess ( http://map.squeak.org/package/812c9d14-5236-4cad-82ea-cc3e3837e30d), and Eliot's simple exemplar REPL thing ( https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/image/LoadReader.... & https://github.com/OpenSmalltalk/opensmalltalk-vm/blob/Cog/image/StartReader... )

I also see that http://www.squeaksource.com/@azqxWT7-o5bjyYS2/vd0deS9q claims to include a REPL package (never tried it, see also http://gulik.pbworks.com/w/page/7760030/REPLServer)

And I'm sure Craig has done something related but I simply cannot get google to find it right now

tim

tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Negligent (adj.), describes a condition in which you absentmindedly answer the door in your nightgown.

On Thu, Nov 7, 2019 at 6:30 AM tim Rowledge tim@rowledge.org wrote:

...

On 2019-11-06, at 7:43 PM, Chris Muller asqueaker@gmail.com wrote: There are two flavors of the Linux VM, you're using the version of the

VM that's a little better, but requires you to install a file into your security limits that allows it to work.

If your VM quits, then you should update it. We removed the exit call after the warning (see David's commit at [1]). We haven't actually observed any misbehavior in case the squeak.conf is not present.

Fabio

[1] https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/32f321583c69ca27e61...

...

Create a text file called "squeak.conf" with these contents (between the

horizontal lines):

...

•   hard    rtprio  2
  

•   soft    rtprio  2
  

The daft thing is that this problem was purportedly fixed in linux kernels as of {mumble-mumble} years ago. Raspbian, for example, does not need it and Raspbian is based on a very conservative branch of Debian

tim

tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Strange OpCodes: IXM: Initiate X-rated error Messages

@Chris:

I believe it has nothing to do with my command. Even if you try to just run

squeak -version in that environment, I think you'll get the same error.

No, this one works:

5.0-201810071412 Mon Oct 8 09:30:27 UTC 2018 gcc 4.8 [Production Spur 64-bit VM] CoInterpreter VMMaker.oscog-eem.2437 uuid: 0e97c106-dd0b-437b-b1aa-e15257288c3f Oct 8 2018 StackToRegisterMappingCogit VMMaker.oscog-eem.2432 uuid: 7b14d114-0e04-4e46-b8a7-4b5e6d87f5fe Oct 8 2018 VM: 201810071412 https://github.com/OpenSmalltalk/opensmalltalk-vm.git ...

(Just wondering why the current Trunk image is delivered with a VM that has version 5.0 ...)

Creating the squeak.conf did not change the behavior for me.

I am using a Ubuntu WSL.

________________________________ Von: Squeak-dev squeak-dev-bounces@lists.squeakfoundation.org im Auftrag von David T. Lewis lewis@mail.msen.com Gesendet: Freitag, 8. November 2019 14:34 Uhr An: The general-purpose Squeak developers list Betreff: Re: [squeak-dev] CommandLineUIManager

On Wed, Nov 06, 2019 at 09:30:44PM -0800, tim Rowledge wrote:

...

On 2019-11-06, at 7:43 PM, Chris Muller asqueaker@gmail.com wrote: There are two flavors of the Linux VM, you're using the version of the VM that's a little better, but requires you to install a file into your security limits that allows it to work.

Create a text file called "squeak.conf" with these contents (between the horizontal lines):

•   hard    rtprio  2
  

•   soft    rtprio  2
  

The daft thing is that this problem was purportedly fixed in linux kernels as of {mumble-mumble} years ago. Raspbian, for example, does not need it and Raspbian is based on a very conservative branch of Debian

Uhmm... it's not a "problem" that needs to be fixed. The Rasbian distribution of Debian Linux is typically intended as single-user system on which you can safely assume that if the user does something dumb, then it serves him/her right when the system locks up.

In the general case of a multi-user system, you don't want someone to elevate their thread priorities and accidentally lock up the entire system. That's why you are not allowed to use real time scheduling priority in ordinary user applications.

Some Linux distributions enforce this by default, and others do not. I think that it more or less depends on the intended audience for the distribution.

Dave